KR100779768B1 - Dynamic and traffic-driven optimization of message routing to geographical addresses - Google Patents

Abstract

Messages are delivered in a communications network by monitoring an arrival rate of messages destined for a geographical area within a short time period. Then, upon reaching a threshold, a multicast group is established for routing the messages to the geographical area. Network devices responsible for delivering the messages join the multicast group and the messages are delivered to the geographical area upon establishment of the multicast group.

Description

DYNAMIC AND TRAFFIC-DRIVEN OPTIMIZATION OF MESSAGE ROUTING TO GEOGRAPHICAL ADDRESSES}

The present invention is used in communication networks to ensure that a service provider can build services to customers or mobile users in a particular selected geographic area.

In the near future, service providers will offer the possibility of customer toe to build services or transfer information there in certain selectable geographic areas. In this way, mobile (wireless) users can be provided with useful services and information related to their current location, for example, specific offers can be advertised to users in the store area.

One of the basic functions required to implement these services is a method of transmitting data from the service provider to selected routers, ie access routers covering these areas via their access technology. This can be accomplished by inserting geographic destination coordinates into each message. It is assumed that every access router knows the coordinates of its coverage area, and all relevant intermediate network systems know the coverage areas of other systems connected to them. In this case, each router tests whether the geographic area covered by its connected access routers or the coverage areas of other routers connected to it matches the target address and matches the result. Send it to the system. The mechanism described above is referred to as GeoCast [1].

The disadvantage of the mechanism described above is that the delay is caused by costly cross checks in intermediate systems before the message can be sent to its destination. In addition, intermediate systems may experience performance bottlenecks that cause congestion if the number of messages to be routed exceeds a certain rate.

Thus, there is a need for a technique that reduces the delay caused by intermediate systems performing cross-checks to deliver a message to a destination within a geographic area, thereby reducing bottlenecks that can cause the system to degrade performance. do.

According to the invention, the problems described above are solved in an efficient and simple manner. The proposed technique allows a message to be delivered to a destination within a geographic area.

The above techniques are achieved by the techniques contained in the independent claims regarding the method and the network apparatus.

The method for delivering messages in a communication network,

The high level router monitors the arrival rate of messages destined for the geographical area within a short time period;

When the arrival rate reaches a threshold, the upper level router establishes a multicast group for routing the messages in the geographic area; network devices forward the messages to join the multicast group. Responsible for doing; And

When the multicast group has been established, the high level router forwarding the messages to the geographic area.

The network device located in the communication network comprises means for using the method according to claims 1 to 7.

Advantages can be seen in the dependent claims that, after the establishment of the multicast group, any further messages destined for the geographic area reached by the upper level router are delivered via the established multicast group and destined for the geographic area. After a predetermined time period in which no further messages arrive, the multicast group is removed and each message destined for the geographic area is determined by a geographic destination address and the geographic location of the messages. The destination address is the same or similar, and monitoring the arrival rate of the message is performed using a soft state message counter, whereby a high speed internet protocol transmission is used to send messages in the multicast group.

The invention will be more fully apparent from the accompanying drawings which are provided for illustration only and not limiting the invention and the detailed description provided below.

1 illustrates routing to geographic regions.

2 is a diagram illustrating dynamically set multicast groups.

3 is a schematic diagram of dynamic multicast group setup.

It is assumed that there is a structure as shown in FIGS. 1 and 2. 1 shows a simple distributed network connected to the Internet via a gateway GW. Access routers in a distributed network have antennas with specific coverage areas. GW and access routers are connected through several intermediate routers. Some of them do not know GeoCast, so each message must be tunneled through IP unicast through them.

The area is defined within the geographical scope. When a message is sent to that area, the message is routed through the distributed network to an antenna that supports that area, that is, access routers whose coverage area is connected to antennas corresponding to that area.

The gateway performs a cross check and sends a message to the two intermediate routers, which have to perform the cross check again until the message reaches the access routers. Next, access routers emit messages in a geographic area.

In the simple example shown in FIG. 1, a total of six cross checks must be performed at the GeoCast nodes of the network.

Simulation results for a typical GeoCast routing system (see [1]) show that the transmission decision in the intermediate system will be made 4,426 more times than the IP router (the fact that the IP router uses firmware should be considered but nevertheless The result shows schematically the scale of the performance difference). These results show that the packet delay resulting from the duration of the transmission decision will increase significantly, and congestion of the intermediate router can occur if the intermediate router has to route too many packets.

The present invention addresses several purposes, such as:

1. The first objective is to find a more efficient solution for routing messages to its goals. That is, to avoid cross checks on the application level or any other application level processing at intermediate routers.

2. An equally important second purpose is to avoid other disadvantages of GeoCast, such as the following: Although multiple GeoCast routers to be addressed will share much of the network path from the transmitting GeoCast router to the receiving GeoCast routers. However, all messages are transmitted via unicast between GeoCast routers, thereby causing unnecessary message duplication in parts of the distributed network. To save resources in a distributed network, it is desirable to replicate the message as close as possible to the point of the network (IP router) where the network paths to different destinations substantially diverge.

Until now, there has been no solution that specifically addresses the dynamic adaptive configuration of distributed networks for routing to geographical addresses. There is no mechanism to utilize dynamic multicast groups to avoid cross checks.

Some efforts have been made to develop mechanisms that can send messages to geographic areas.

GeoCast :

The GeoCast mechanism, introduced at the beginning of this document, relies on cross-checks of the geographical target address of the message through the coverage area of each intermediate system. As already explained, the test requires so much computational power and time that it cannot be ignored. In addition, costly address resolution is required according to the representation of the address in the message. Cross checks should be performed for each single message. At the very least, GeoCast uses cache entries to speed up transmission decisions. Messages are identified either via ID or through other values that are part of the message header, such as source and destination addresses. After the first packet is received, the final transmission decision is stored in the cache. Subsequent messages with the same ID may omit cross checking, but nevertheless the message must be processed by the geo routing module to determine the ID and corresponding cache entry at each router. An additional disadvantage is that messages are sent over unicast and therefore there is no mechanism in place (eg no multicast routing) to reduce the overall amount of messages to be sent in the network. There is no mechanism for dynamic optimization of message routing.

Placement of static multicast groups:

Another mechanism that relies on IP multicast is called the GPS-multicast routing scheme [2]. To speed up the transmission decision, static multicast groups are set up in a distributed network. Routers are coupled to Atoms and a multicast address is assigned to them. Several atoms can be recombined into partitions with unique addresses, which can be combined into larger partitions. Next, the geographical target address polygon of the message is approximated through the smallest partition that contains it and sent to the corresponding IP multicast address. This requires mapping from geographic destination addresses to multicast groups. One drawback of this solution is the fact that multicast groups will rarely match the exact target area. This means that some systems in the network will receive packets erroneously, perform a cross check and discard them. An important disadvantage of this solution is that it is necessary to divide the coverage areas into terms and compartments before a static multicast mechanism can be utilized in the network. This leads to a difficult network planning problem that requires a lot of administrative interaction for the preconfiguration of the network during its operation with costly considerations or estimates of traffic patterns (to date not yet known). Another problem with this solution is that it must be determined a priori how the geographical topology is divided into partitions and which multicast groups will be created, which requires (rare) multicast addresses to be assigned and signaling traffic exchanged. It is very easy to derive multicast groups that are only rarely or never used to be handled.

Placement of dynamic multicast groups on the last hop:

In order to distribute the message between the access router and mobile clients in its coverage area, the GPS-multicast routing scheme [2] deploys multicast groups for "last mile" routing. The access router assigns a group to all mobile clients in a particular area. This also occurs dynamically, for example based on specific geographic polygons. All mobile clients in the polygon can join the group because they know their geographic address derived from their assumed GPS module. In contrast to the invention described in the report of the present invention, multicast groups are not deployed in a distributed network to reduce packet delays in intermediate systems to speed up transmission decisions, but groups are located between access routers and mobile clients. It is only valid for the purpose and aims to reduce the amount of unnecessary messages that mobile clients will receive and to save the rare resources of the air interface. Another disadvantage of this mechanism is that one multicast group is allocated and maintained for each particular individual geographic address.

In order to reduce transmission delay and the required amount of cross checks and messages, the invention described in the present invention report provides a mechanism for establishing multicast groups in a network that are dynamically adapted to the data traffic that occurs. If a certain amount of messages arrive in a short time with the same or similar geographic target address, the IP multicast group is cross-checking, requiring all the access routers responsible for sending the message, that is, the transmission procedure over the distributed network. When the message is received later it will generally be set up by the network containing all such access routers to send the message. The disadvantage of the described mechanism is based on the fact that the IP multicast transmission decision in the intermediate system is made many times faster than the GeoCast system.

Specifically, monitors when a high level router, such as a gateway, in a distributed network receives several messages with the same or very similar destination addresses in a short time (e.g., service providers are responsible for users in an area with a particular event). If you want to address it). The duration or arrival rate is chosen in such a way that the router can estimate that multiple messages with the same address will follow. This can be implemented, for example, via a soft status message counter. By monitoring the traffic load of the geographic area, the GeoCast router can calculate the optimal geographic area for the multicast distribution of geographic messages, thereby making the number of various tradeoffs / multicast groups unnecessarily Optimize the number of distributed messages, signaling load for multicast group maintenance, and so on.

Thereafter, access routers responsible for message delivery to part of the addressed geographic area are requested to join the dynamically generated multicast group. The request is sent from the upper level router to the access routers via a standard geographic addressed message and intercepted by the access routers. After each access router processes the request, joins the multicast group and verifies it, messages to the area are sent directly to the IP multicast group. All intermediate systems utilize high-speed standard IP transmission of messages to eliminate the need for costly cross checks in the intermediate systems, and allow the network to replicate messages as close to the target systems as possible. Access routers remove the multicast IP information from the messages and send the messages according to the geographical target addresses of the messages. The multicast group is removed if no message with respective target addresses arrives anymore for a certain time.

The proposed mechanism is then suitable for creating a service, which services are related to specific areas (so-called area based services). This is a fundamental technique that can be deployed in the network infrastructure to speed up transmission decisions in intermediate systems. Through faster decisions, the delay in message delivery is also shortened.

2 illustrates a distributed network with dynamically configured multicast groups.

The following section provides a more detailed description of the invention.

3 provides a schematic diagram of a dynamic multicast group establishment procedure in a distributed network. The description is as follows:

1. A router in a distributed network proximate the gateway or gateway (ie, the upper level router) itself has a soft status message counter. The soft state message counter counts GeoCast messages with the same or similar geographic target address that appear within a certain time period. Unless any dynamic multicast groups are established, messages are routed through GeoCast (ie, cross checks or other application layers involving decision) to other access routers to access routers that emit messages directly in a geographic range. Is sent via).

2. After a certain amount of messages addressed to a particular area arrives within a specified time period, the upper level router begins creating a temporary multicast group to speed up the transmission procedure in the distributed network. Therefore, the upper level router encapsulates the Request to Join temporary group message in a GeoCast message. This message contains the address of the temporary multicast group. Similarly, other messages prior to these messages are sent to access routers via GeoCast via the distributed network.

3. After receiving the encapsulated message, access routers begin processing the Request to Join message. The access routers add a (*, AR-G) state indicating that they have received multicast messages sent by arbitrary sources to the multicast group address sent in the original message from the upper level router.

4. When the state is added, a Join temporary group message is sent to the upper level router.

5. For later use of the multicast address, the upper level router must ensure that all relevant access routers join the multicast group. Therefore, delivery of exchanged join messages must be reliable. This means that each intermediate router sending an encapsulated Request to Join temporary group message to another router must receive a Join temporary group message from it, otherwise the process will be repeated.

6. After the upper level router receives the Joint temporary group messages, the router adds the appropriate multicast state to the unique routing entries.

7. If a GeoCast message with matching destination addresses arrives at the top level router, the message is encapsulated in an IP multicast packet and sent directly to the access routers. Transmission in a distributed network is made through fast IP routing decisions.

8. The upper level router monitors incoming messages to other geographical destination addresses. If other messages appear to be sent to the same or very similar geographical addresses, they will be sent to the same multicast address.

9. If no message arrives with a geographic destination address during a particular time period, the dynamic multicast group is removed from the network. This can be accomplished through explicit pruning message or timeouts.

The following important characteristics are achieved:

Multiple IP multicast groups for different geographic destination addresses do not consume signaling to maintain each multicast group for a geographic area that consumes rare IP multicast addresses and receives little traffic. Can be maintained.

The computational time required for transmission decisions at intermediate routers in the destination network is reduced, thereby enabling greater volume of traffic on that network.

Since transmission decisions for multicast messages are obtained only by looking at the IP address of those messages, packet delay is significantly reduced in a distributed network.

Traffic volume in the network is reduced by avoiding unnecessary transmission of duplicate messages along the shared portions of the network patch (leveraging such an important feature of multicast routing for the distribution of GeoCast messages.

• Routing optimization processing is traffic based and thus can be implemented as economically as possible with regard to the exchange of signaling messages and the assignment of addresses.

In a distributed network, multicast groups are set up dynamically to reduce transmission delay and reduce the amount of duplicate messages.

Claimed that new methods have been developed for:

To avoid costly application level transmission decisions in the intermediate network of the distributed network for routing to geographic addresses,

To optimize routing in distributed networks in a traffic-based fashion,

To allow dynamic calculation of "high demand geographical areas" and to create multicast groups specifically for efficient message distribution to such areas,

To avoid difficult and cumbersome "a priori partitioning phase of geographic regions" for the construction of static multicast groups.

In summary, the present invention represents an important step in implementing an efficient distributed network that is self-configured for routing messages to geographic addresses.

The following shows an example for the description of the present invention. It relates to the structure of a sample access network having a distributed network of specific carriers. The network is in this case connected via a gateway to the Internet, which is a high level router. Assume that a specific public event (eg, a soccer game) takes place in the coverage area (example) of access routers connected to a network.

Now a company wants to advertise its products in the area during the event. The company determines the geographical coordinates of the area and inserts the determined coordinates into the advertising message. Next, the message is sent to the gateway of the distributed network (generally, this step requires interaction of the service broker but this is outside the interest of the present invention).

Advertising messages are now sent via GeoCast via a distributed network, and are emitted once by the products, but send messages in the same area at short time intervals.

After some messages, the gateway notices that this geographic address is very often used. It constructs a Request to Join temporary group message and inserts an assigned temporary multicast address. This message is encapsulated in a specific GeoCast message with the same geographical coordinates as other messages. Thereafter, the message is sent to the distributed network.

Access routers receive messages and decapsulate the messages. Instead of delivering the messages to the area, access routers start processing the Request to Join message by joining a multicast group having a predetermined multicast address.

Thereafter, the Join temporary group message is sent to the gateway. When intermediate systems and the gateway receive all Request to Join messages (which means that all affected access routers have received the message from the gateway), the gateway adds the appropriate multicast state to the unique routing entries. .

If a GeoCast message with a matching destination address reaches a top level router, the message is encapsulated in an IP multicast packet and sent directly to access routers. Transmission in a distributed network is made through fast IP routing decisions.

Access routers decapsulate the messages and distribute the messages to the area via their antennas.

Other companies also want to send advertisements in the area of public events. The company determines geographic coordinates that are only slightly different from the geographic coordinates used by the first company. The company also sends messages with coordinates (ie, similar geographical addresses) to the gateway. The gateway recognizes the similarity and sends messages to the same multicast group address that has already been set up. The rest of the message processing is the same as described above.

When the event ends, both the first company and the second company no longer send messages to the area. After some time, the temporary multicast address is removed from the distributed network, for example through timeouts or explicit pruning messages.

Although the present invention has been described through the preferred embodiments described herein, those skilled in the art will recognize other embodiments and modifications that may be made without departing from the scope of the present invention. All such modifications are intended to be included within the scope of the claims appended hereto.

Reference:

[1] T. Imielinski, J. Navas, "GeoCast-Geographic Addressing and Routing", Proceedings of the Third ACM / IEEE International Conference on Mobile Computing and Networking (MobiCom'97), Budapest, Hungary. September 1997.

[2] T. Imielinski, J. Navas, "GPS-Based Addressing and Routing" IETF Request for Comments 2009, rfc2009.txt, November 1996.

Claims (9)

A method for delivering messages in a communication network, comprising: Monitoring, by the upper level router, the arrival rate of messages destined for the geographic area within a short time period; When the arrival rate reaches a threshold, the upper level router establishing a multicast group for routing the messages to the geographic area, wherein network devices responsible for forwarding the messages to the multicast group. Joined-; And And when the multicast group is established, the high level router forwarding the messages to the geographic area. 2. The method of claim 1, further comprising forwarding, via the multicast group, any further messages destined for the geographic area reached by the upper level router after the multicast group is established. Delivery method. 3. The method of claim 1 or 2, further comprising the step of removing, by the upper level router, the multicast group after a predetermined time period has elapsed where no further messages destined for the geographic area have arrived. Including the messages. The method of claim 1 or 2, wherein the messages destined for the geographic area are each determined by a geographic destination address. 5. The method of claim 4, wherein the geographical destination address of the messages is the same or similar. 3. The method of claim 1 or claim 2, wherein said arrival rate monitoring step is performed using a soft status message counter. 3. The method of claim 1 or claim 2, wherein high speed internet protocol transmission is used to send messages in a multicast group. A network device located in a communication network, A network device comprising means for using the method according to claim 1. 10. The network device of claim 8, wherein the network device is a router or a gateway.

Images